-2 0 2 4 -6 -4 -2 0 2 4 Mean Refrac/ve Error Di ffer en ce i n R efr ac/v e E rr or (W ASS pe dia tr ic -S R) -2 0 2 4 -6 -4 -2 0 2 4 Mean Refrac/ve Error Di ffer en ce i n R efr ac/v e E rr or (W ASS adult -S R) -2 0 2 4 -6 -4 -2 0 2 4 Mean Refrac/ve Error Di ffer en ce i n R efr ac/v e E rr or (W ASS adult -W AS S pe dia tr ic )
Significance
:
•Clinically, refractive error degrades visual acuity in the feline
veterinary patient. The cat is an important historical and current model of ophthalmic disease, especially disorders of refractive error.
•Refraction by streak retinoscopy, while the gold standard,
requires advanced training and practice.
•Objective autorefraction, such as with the Welch Allyn
SureSight™ autorefractor, could offer a faster, more practical, and accurate method of refraction in the cat.
Hypothesis
:
Refractive error as measured by automated vs. manual methods in the domestic cat will not significantly differ.
AUTOMATED VERSUS MANUAL REFRACTIVE ERROR MEASUREMENTS IN DOMESTIC CATS
AM Cleymaet,
1EN Harb,
2AM Hess,
3and KS Freeman
11. Department of Clinical Sciences, College of Veterinary Medicine, Colorado State University 2. Department of Vision Science, University of California Berkeley, School of Optometry 3. Department of Statistics, College of Natural Sciences, Colorado State University
2.
Is there good agreement between
methods?
Results: 1.
What degree of refractive
error is present?
Conclusions:
Routine measurement of refractive error by streak
retinoscopy in more than just the horizontal and
vertical meridians should be further investigated
in clinical veterinary medicine.
While there was a significant difference between
methods, the level of agreement between SR and
WASS
adultfor measurement of refractive error in
the adult domestic cat is good. For WASS, adult
setting is recommended for clinical use.
4.
Which setting is recommended for
the Welch Allyn SureSight™
autorefractor?
3.
Does the method of refraction
change the diagnosis for the overall
refractive state of the eye or patient?
Objective
:
To compare the results of streak retinoscopy (SR) vs. the Welch Allyn SureSight™ autorefractor (WASS) in normal cats and
determine the appropriate WASS setting (WASSadult vs.
WASSpediatric) for use in the domestic cat.
n SE (D) Astigmatism (D) Astigmatism <1D (%) (% eyes)WTR (% eyes)ATR (% eyes)Oblique SR 60 eyes
30 cats +1.05 ± 0.97 -0.47 ± 0.59 73.33 (eyes) 66.67 (cats) tendency toward ATR
WASS SEadult 56 eyes 28 cats +0.60 ± 1.15 -0.75 ± 0.58 66.07 (eyes) 42.86 (cats) 35.71 37.50 26.79 SR - same
cohort +1.00 ± 0.99 -0.44 ± 0.59 71.43 (eyes)67.86 (cats) tendency toward ATR
WASS SEpediatric 16 eyes 8 cats +2.75 ± 0.98 -0.88 ± 0.35 50.00 (eyes) 37.50 (cats) 25.00 25.00 50.00 SR - same
cohort +1.35 ± 0.93 -1.08 ± 0.69 31.25 (eyes)12.5 (cats) tendency toward ATR
The difference between methods was statistically significant for WASSadult vs. SR (p ≤ 0.001, n=56 eyes), WASSpediatric vs. SR (p =
0.01, n=16 eyes), and WASS_adult vs. WASSpediatric (p ≤ 0.001, n=12
eyes). For both comparisons (WASSadult vs. SR and WASSpediatric vs.
SR), there was not a significant difference between eyes (p > 0.05) and there were no significant sex differences (p > 0.05).
Significant correlation exists between refractive error
measurements by WASS
adultand SR. WASS
adultresults in
a greater proportion of DSE values centered around 0.
Yes, there was
good agreement
between
WASS
adultvs. SR.
A
(A) Histogram of distribution of refractive error values (per eye) in terms of sph. eq. as obtained with different refractive techniques. Fitted (Gaussian) distribution lines superimposed. -2 0 2 4 0 5 10 15 20 Spherical Equivalent (D) Num be r of E ye s Pediatric SE_SR Pediatric SE_WASS Adult SE_SR Adult SE_WASSC
D
Mean vs. difference plot of (C) WASSadult and SR.(D) WASSpediatric and SR.
(E) WASSadult and WASSpediatric . Mean
difference (dashed line) and 95% limits of
agreement (dotted lines) are depicted. The 95% limits of agreement for
WASSadult
vs. SR was (-1.80 D, +0.99 D),
WASSpediatric vs. SR was (-0.75 D, +3.55 D), and WASSadult vs. WASSpediatric was (-4.88 D, +0.34 D).
WA
SS
SR
Total=28 hyperopic myopic emmetropic anisometropicBi
(B) Distribution of overall diagnosis of refractive error in a given eye based on sph. eq. obtained with different refraction techniques. (Bi) SR vs WASSpediatric (n = 16) (Bii) SR vs WASSadult (n = 56)Ci
Bii
Cii
(C) Distribution of overall diagnosis of refractive error in a given cat based on sph. eq. obtained with different refraction techniques. (Ci) SR vs WASSpediatric (n = 8) (Cii) SR vs WASSadult (n = 28)Yes, the method of refraction can change the diagnosis
for the eye or patient’s overall refractive state.
A
(A) Correlation between DSE and the reliability number reported by the WASSadult. Line depicts the x = y line of equality. There was a significant correlation between DSE and the WASSadult reliability number (p = 0.04, r = -0.28) but (data not shown) not WASSpediatric (p = 0.81, r = -0.0.7) 0 5 0 5 Reliability Number DS E ( D)Given the above and that there was an intermethod
discrepancy of 0.41 D and variability of 1.40 D for
WASS
adultvs. SR, WASS
adultis the recommended setting
for clinical use of the WASS refractometer.
Funding: Supported in part by the Center for
Companion Animal Studies at Colorado State University.
Discussion:
•
Comparison to prior studies: A report indicated that
outdoor cats have mean refractive error of +1.15±0.18
D and indoor cats -0.81±0.20 D; cats were refracted at
each meridian. (Belkin et al. 1977) A more recent study
reported a refractive error of –0.78 ± 1.37 D, but cats
were strictly refracted along the horizontal meridian.
(Konrade et al. 2012)
•
SR is routinely performed in clinical veterinary
medicine along the horizontal and vertical meridians.
•
In human optometry and ophthalmology, SR is
performed along the principal meridians.
•
Given the results of the present study, there may be
more oblique astigmatism in the domestic cat than
assumed.
•
Limitation: there was a greater range of refractive
errors in the WASS
adultvs SR cohort, including myopic
eyes, and this may result in data skew.
Methods:
•Refractive error determined in 30 young adult domestic short
haired cats (60 eyes) with normal, non-cyclopleged eyes via SR in the horizontal and vertical meridians.
•Refractive error also determined via WASSadult (n=28 cats (56
eyes)) and WASSpediatric (n=8 cats (16 eyes)).
•Refractive error determined by both WASSadult and WASSpediatric
in 6 cats (12 eyes).
•Animals handled in compliance with guidelines of CSU’s
Institutional Animal Care and Use Committee. All procedures carried out according to ARVO Statement for the Use of Animals in Ophthalmic and Vision Research.
Analysis:
•Limits of Agreement: Calculated as outlined in Bland & Altman
2007 - “Method where the true value varies”. Maximal inter-method discrepancy of 0.50 D and variability of 2.30 D considered a priori to be clinically acceptable based on prior literature. (Bonds 1974, Sivagurunathan 2011, Paff et al. 2010; Akil et al. 2015; Prabakaran et al. 2009)
•Difference between methods: Random effects model fit using
lme4 package. Analysis done separately for adult or pediatric
WASS values. Response variable was difference between
methods (WASS - SR). Cat included as a random effect to account for multiple observations (2 eyes) on most cats.
•Data presented as mean ± SD.
(A) Correlation of the refractive error
between SR and WASSadult (red points) and between SR and WASSpediatric (blue points). Line depicts the x = y line of equality. (Ai) There was a significant correlation between WASSadult and SR (r = 0.80, p < 0.0001). There was no significant correlation between
WASSpediatric and SR (r =0.38, p = 0.14)
(Aii) Plot of those cats (n=6) for which
refraction was performed by all 3
methods. (B) Histogram of frequency distribution of the difference in
spherical equivalent / sph. eq. (DSE) between WASSadult vs. SR and
WASSpediatric vs. SR. Fitted (Gaussian) distribution lines are superimposed.
